HUBU PaddlePaddle Developer GroupHUBU · PADDLEPADDLE DEVELOPER GROUP Manchester Metropolitan Joint Institute joint university logo PaddlePaddle
PEOPLE

People

Different disciplines, connected through shared learning and practice.

About the group Our community

The HUBU PaddlePaddle Developer Group is a university community for AI learning and hands-on projects. Students and teachers use PaddlePaddle resources to collaborate through technical exchange, open-source work and campus competitions.

Students in electronic information engineering, software engineering, materials science and other disciplines share ideas at training camps and discover connections between AI and their own fields.

College report: UP Camp industry visit ↗

Faculty advisors Mentors

Portrait of Riji Yu
Photo provided by the team

Riji Yu

School of Art and Design, Hubei University

Professor · Doctoral Supervisor · Associate Dean

Researches virtual reality and interactive media, including visualization, digital twins and cultural heritage digitization, bringing digital technology into cultural experiences and creative design.

Research & academic experience

PhD in Digital Media from Wuhan University; Director of Hubei University's Culture and Technology Integration Innovation Research Center.

Research spans VR/AR/MR, digital 3D, interactive media, digital entertainment, cultural heritage preservation and digital tourism.

Work includes digital displays of traditional Tujia villages and a metaverse platform for the Marquis Yi of Zeng chime bells. The Cultural Heritage XR Immersive Interaction System won Best XR Application at the 2024 SIVA Awards.

Full profile
Official portrait of Bo Ye
Photo source ↗

Bo Ye

School of Artificial Intelligence · Information and Communication Engineering

Associate Professor · Master's Supervisor

Researches miniature intelligent equipment and automation, including closed-loop MEMS control, big data and AI, capsule-endoscope control and medical image analysis.

Research & academic experience

Bachelor's in Electronic Information Engineering from Hubei University; Master's in Communication and Information Systems and PhD in Mechanical Manufacturing and Automation from HUST.

Leads research on closed-loop capsule-endoscope control and magnetic spiral capsule localization, with papers in IEEE TASE, IEEE TIM and IEEE Sensors Journal.

Received third prizes in the 2024 Hubei Science and Technology Progress Awards and the 2017 Hubei Technological Invention Awards.

Full profile
Official photograph of Dan Zhang
Photo source ↗

Dan Zhang

School of Artificial Intelligence, Hubei University

Associate Professor

Researches signal processing, the Internet of Things and intelligent instruments, drawing on experience in electronic product development and hardware team management.

Research & academic experience

Graduated from Hubei University in Applied Electronic Technology and earned a master’s degree in Communication and Information Systems from Huazhong University of Science and Technology in 2004.

Previously an application engineer at Wuhan P&S Information Technology and a hardware department manager at Wuhan Bit Renjie Technology; now an associate professor at Hubei University’s School of Artificial Intelligence.

Has undertaken research supported by the Natural Science Foundation of Hubei Province and industry partnerships.

Official university profile
Official portrait of Tingting Zhang
Photo source ↗

Tingting Zhang

School of Intelligent Manufacturing, Hubei University

Lecturer · Master's Supervisor

Researches intelligent agricultural machinery, precision agriculture, agricultural robotics and biomass carbon reduction at the intersection of intelligent manufacturing and modern agriculture.

Research & academic experience

Previously a lecturer and postdoctoral researcher in agricultural engineering at Shandong Agricultural University; joined Hubei University's School of Intelligent Manufacturing in June 2025.

Leads a 2026 Hubei Natural Science Foundation Young Scientists Category B project on soil adhesion at agricultural soil–tool interfaces.

Leads an AI + OBE teaching reform project in intelligent manufacturing; her team won second prize in the Second Hubei Agricultural Machinery Innovation Design Competition.

Full profile

People Team members

Students from different disciplines, growing together through research and engineering.

Zhenxin Lin

Zhenxin Lin Zhenxin Lin

Team Leader

2023 entry · Electronic Information Engineering · Joint Institute

Interested in LLM pretraining and post-training, reinforcement learning, multi-agent decision-making, and AI for vision and medicine. Following algorithm research internships at China Telecom and Z.ai, he currently works on foundation models and game policies at Kingsoft's AI Product Center.

Since 2024, he has helped establish and run the HUBU PaddlePaddle developer community, organizing talks, workshops, competitions and camps that connect campus learning with real projects.

LLM TrainingReinforcement LearningMulti-agent SystemsComputer Vision
Research & practical experience

Research experience

Endoscopic lesion detection in degraded target frames: bidirectional trajectory query recovery
Research paper · Researcher and first author

  • Problem: blur, low light, specular reflections and occlusion degrade video frames and cause missed endoscopic lesion detections.
  • Contribution: BiTraQ-DN restores missing detection candidates through trajectory prediction, visual-memory validation and bidirectional box consistency. Reported FROC-AUC@1FP is 0.1559 on Kvasir-Capsule and 0.3714 on LDPolypVideo, with 43.97 frames per second on an H100 GPU.
  • Under review at npj Digital Medicine (2025 impact factor: 18.0; five-year impact factor: 19.7)

CardKS: Residual key-structure modeling for long-horizon decisions
Research paper · Researcher and first author

  • Problem: hidden states, combinatorial action spaces and long planning horizons constrain decisions in imperfect-information games.
  • Contribution: CardKS combines residual-structure retrieval with candidate-conditioned policy–value learning; KSCB and KSPlay support evaluation and scalable training. Reported win rates against representative learned baselines are 71.5%, 54.0% and 58.37% across three games, with simulation throughput up to 2.10 times the baseline.
  • Under review at AAAI 2027 (CCF A)

Multi-agent policy-factor evolution: interpretable adaptation across imperfect-information games
Research paper · Researcher and co-first author

  • Problem: existing policies are difficult to explain, reuse and adapt across imperfect-information games.
  • Contribution: policy knowledge is compiled into executable factors, enabling adaptation without retraining. Across Guandan and Gin Rummy, all nine tested rule-based, RL and language-model policies improved by 3.33–50.00 percentage points.
  • Under review at AAAI 2027 (CCF A)

Safety-filtered distributed Koopman model predictive control
Research paper · Researcher and second author (advisor is first author)

  • Problem: packet loss can remove both neighboring predictions and collision constraints from distributed model predictive control.
  • Contribution: a local-sensing, acceleration-level quadratic-program safety filter separates Koopman-MPC planning from safety assurance. With static A* waypoints, all 20 trials were collision-free and all 80 goals were reached; the average peak pairwise constraint slack fell by 62.4%.
  • Under review at AAAI 2027 (CCF A)

Industry internships

  • Kingsoft · AI Product Center| March 2026–present
    Contributes to dense LLM pretraining, distributed training on Ascend NPU clusters, long-context extension and CardKS policy-model development.
  • Z.ai · Business Intelligence Division|2025.09—2026.02
    Worked on an automotive manual question-answering system, including dataset construction, GLM supervised fine-tuning and inference deployment with vLLM and SGLang.
  • China Telecom Hubei · AI R&D Department|2025.06—2025.09
    Worked on urban public-safety visual analysis, combining YOLO, U-Net, Qwen-VL fine-tuning and retrieval-augmented generation in multimodal applications.

Open source & applications

  • DanKS ↗: a reinforcement-learning research system for imperfect-information games, covering training, evaluation and cross-platform deployment.
  • awesome-latex-skills ↗: academic writing workflows for agents, covering compilation repair, language editing, publication-format migration and paper reading.
  • AI-assisted table tennis training system: uses PaddlePaddle and ERNIE to analyze movement and performance and generate training feedback; demonstrated at Baidu World 2025.

Competitions & campus experience

  • 2024: First Prize in the national finals of the National 3D Digital Innovation Design Competition.
  • 2024: National Bronze Award, 14th Challenge Cup Chinese College Students Entrepreneurship Plan Competition.
  • 2026: Meritorious Winner (M Award), Mathematical Contest in Modeling.
Dongyang Yu

Dongyang Yu Dongyang Yu

People

2023 entry · Software Engineering · Joint Institute

Interested in graph neural network test-time adaptation, LLM applications and multi-agent collaboration. First author of SETTA, applicant on two Chinese invention patents, and lead of a provincial undergraduate innovation and entrepreneurship project.

Graph Neural NetworksTest-time AdaptationMulti-agent Systems
Research & practical experience

Selected paper

SETTA: Parameter-Free Test-Time Adaptation for Graph Neural Networks via Spectral-Energy-Guided Semantic Refinement.
Dongyang Yu, Xia Cui, Rong Xiao.
Big Data and Cognitive Computing, 2026, 10(8), 260.

Projects & patent applications

  • Project lead of Wudang Lingshu, a multi-agent smart culture and tourism platform supported by a provincial undergraduate innovation program.
  • Baduanjin AI Coach: movement assessment combining pose estimation and large language models.
  • Multi-agent collaborative big-data management system and method: first inventor; application published.
  • DeepResearch-based information retrieval and generation method and system: third inventor.

Practice & exchange

  • LLM application development intern at the Hunan University–China Mobile Joint Institute for Industrial Intelligence, working on robot dialogue flows and emotional decision-making.
  • Joined a summer program at Kellogg College, University of Oxford, through Hubei's 2025 Outstanding Undergraduate Overseas Study Program.
  • 2026: Meritorious Winner (M Award), Mathematical Contest in Modeling.
Junyan Liu

Junyan Liu Junyan Liu

People

2023 entry · Electronic Information Engineering · Hubei University

Focuses on project operations, market collaboration and electrical hardware products, with experience in marketing internships, industry liaison and energy-sector fieldwork.

Project OperationsIndustry CollaborationElectrical Hardware
Operations & project experience
  • Interned with the marketing operations department at New Oriental’s Jinan campus and the marketing and sales department at TGOOD, contributing to marketing operations and industry project liaison.
  • Led Hubei University’s Dian Jing Zhi Bi student team in 2024, organizing energy-sector fieldwork and exchanges with Zhunneng Group.
Junwei Chen

Junwei Chen Junwei Chen

People

2023 entry · Electronic Information Engineering · Hubei University

Focuses on VR/XR interaction, digital twins and large language model applications. Contributed to the Xiaoyao Wudang location-based VR project, with experience in 3D modeling, model fine-tuning and deployment.

VR / XRDigital TwinLLM Applications
Research & practical experience

Technical development & projects

  • Xiaoyao Wudang location-based VR: contributed to technical development. The project has been deployed at Wuhan MixC, Wuhan Binjiang Paradise Walk and Optics Valley K11.
  • XR education: worked on projects including a digital museum of the Marquis Yi of Zeng’s chime bells and an acupuncture training system, applying 3D modeling and mixed reality.

Industry internships

Gained technical development experience at Wuhan Shuwen Technology and Hubei Digital Culture and Tourism Technology Group. Worked on data processing, model fine-tuning and application deployment during internships at China Telecom and Z.ai.

Competition experience

  • National First Prize, 17th National 3D Digital Innovation Design Competition.
  • National Bronze Award, 14th Challenge Cup China College Students Entrepreneurship Competition, for the Kejiao Yuanlv project.
  • Hubei Division Second Prize, 13th Future Designer competition.
Haowen Xiao

Haowen Xiao Haowen Xiao

People

2024 entry · Electronic Information Engineering · Hubei University

Interested in computer vision, low-light image enhancement and edge AI, with experience in robot reproduction, smart buildings and drone maintenance projects. Works on model deployment, device communication and hardware–software integration. A 2026 MCM Meritorious Winner.

Computer VisionImage EnhancementEdge AIRobotics
Research & practical experience

Research exploration

Low-light image enhancement: helped adjust model architecture, loss functions and training configurations; organized iSAID data and training, validation and low-light pipelines; configured the GPU environment and validated forward passes; explored frequency-domain processing and mixture-of-experts models.

Hands-on projects

  • MiniDuck bipedal robot reproduction: assembled the robot, configured Raspberry Pi, deployed an ONNX gait policy, calibrated joint zero positions and tuned control parameters to achieve stable standing and walking.
  • Smart-building edge AI system: collaborated with UK students in the 2025 MMU summer camp, handling MQTT communication and integration; deployed face recognition to Raspberry Pi 5 and connected cameras and control feedback.
  • Integrated intelligent substation maintenance platform: helped develop and debug the inspection-drone subsystem, prepare project materials and present the work; the project received a university-level Gold Award in the China International College Students' Innovation Competition.

Industry internships

Baidu AI Technology Department: helped update and repair PaddlePaddle teaching projects, resolving dependencies, data paths and model-interface compatibility, and contributed to an agricultural medical consultation assistant and an appliance fault-detection agent.

Competition experience

2026 MCM Meritorious Winner (M Award), responsible for data cleaning, machine-learning training, results preparation and English paper writing.

Manbei Li

Manbei Li Manbei Li

People

2024 entry · Electronic Information Engineering · Joint Institute

Focused on robot control, 3D visual perception and mapping between physical systems and digital twins. Contributed to 3D modeling, PLC and HMI interaction, and voice-interface development, earning a national first prize in the Industrial Collaborative Robotics and Digital Twin Technology Innovation Application Competition.

Robot Control3D VisionDigital Twin
Competitions, projects & student leadership

Competitions & projects

  • 2026: National Finals First Prize, Industrial Collaborative Robotics and Digital Twin Technology Innovation Competition, part of the Chinese College Students Mechanical Engineering Innovation and Creativity Competition.
  • Third prize for the Smart Building project at the Institute’s 2026 Academic Summer Camp.
  • Second prize for the Smart Ship project at the Second UP Camp, with recognition as an Outstanding Camper.
  • Third prize in the Institute’s speech competition for UK liaison representatives.

Study & student leadership

  • Former head of the New Media section of the Institute Youth League Committee’s Publicity Department; contributed to publishing news on the university news website.
  • Received the first-class Outstanding Scholarship and Academic Excellence Scholarship for 2024–2025, along with Three-Good Student and Outstanding Student Leader honors.
  • Represented the Institute at the sixth United Nations agency presentation event and served as a bilingual volunteer at the 2026 World Youth Development Forum.
  • A member of the Institute’s Manwu Dance Troupe; received university-level recognition with the team at Hubei University’s 37th Dance Competition.
Jiatong Song

Jiatong Song Jiatong Song

People

2024 entry · Electronic Information Engineering · Joint Institute

Works on mathematical modeling, industrial collaborative robots and digital twins. Received an MCM Honorable Mention and a national first prize in industrial collaborative robotics and digital-twin innovation. Serves as Deputy Secretary of the Institute's Youth League Committee and participates in student organizations and innovation activities.

Mathematical ModelingRoboticsDigital Twin
Competitions, projects & student leadership

Competitions & projects

  • 2026: Honorable Mention (H Award), Mathematical Contest in Modeling.
  • 2026: National Finals First Prize, Industrial Collaborative Robotics and Digital Twin Technology Innovation Competition, part of the Chinese College Students Mechanical Engineering Innovation and Creativity Competition.
  • 2026: Grand Prize, Institute Academic Summer Camp.
  • 2025: Third Prize, Hubei University AI Cup Electronic Design Competition, for an image-recognition-based humidifier water-level monitoring and automatic refill system.
  • 2025: Outstanding Participant, Second UP Camp.
  • An invention patent application for an intelligent water aeration pump was accepted in August 2026.

Study & student leadership

  • Deputy Secretary of the Institute's Youth League Committee, contributing to student organization work.
  • 2024–2025: Second-Class Outstanding Student Scholarship and Merit Student recognition.
  • 2025: Outstanding Youth League Member; led the class to receive Five-Four Red Flag Youth League Branch recognition.
  • 2024: Excellence Award in Hubei University's Most Beautiful Study Notes undergraduate competition.
Xinyue Zhang

Xinyue Zhang Xinyue Zhang

People

2024 entry · Electronic Information Engineering · Joint Institute

Takes part in PaddlePaddle learning and technical exchanges, contributes programming work to the WisdomScape smart campus project, and gains practical experience through projects and competitions in electronic design, robotics, digital twins and embedded systems.

Electronic DesignRoboticsDigital TwinEmbedded Systems
Projects & awards

Project experience

  • Contributed programming work to the WisdomScape smart campus project.
  • Participated in a student innovation and entrepreneurship project approved at the provincial level.

Competition achievements

  • First Prize, Southern Region, Information Technology Frontiers special competition of the National Undergraduate Electronic Design Contest.
  • Provincial First Prize, Industrial Collaborative Robotics and Digital Twin Technology Innovation and Application Competition.
  • Provincial Second Prize in Embedded Chip and System Design.
  • National Second Prize, China Artificial Intelligence and Robotics Competition.
Zijing Huang

Zijing Huang Zijing Huang

People

2024 cohort · Materials, Class 2401 · Hubei University

Participates in a national-level undergraduate innovation project and leads a provincial-level project. Awards include Second Prize in the Central South Regional Contest of the Chinese Collegiate Computing Competition, a silver award in a synthetic biology innovation competition, and a university scholarship. These interdisciplinary projects provide hands-on innovation experience.

MaterialsInterdisciplinary Projects
Projects & awards
  • Participates in one national-level undergraduate innovation project and leads one provincial-level project.
  • “Wenjing Chengying,” a multimodal AI platform for generating comic videos from Chinese novels, received Second Prize in the Central South Regional Contest of the 19th Chinese Collegiate Computing Competition (2026).
  • Received a silver award in a synthetic biology innovation competition and a university scholarship.
Xuan Zhao

Xuan Zhao Xuan Zhao

People

2024 cohort · Software Engineering · Hubei University

Follows advances in artificial intelligence and actively participates in PaddlePaddle technical exchanges and innovation discussions. Contributes to smart building, ship agent and digital twin projects, exploring practical applications at the intersection of software engineering and intelligent technologies.

Artificial IntelligenceSoftware EngineeringDigital Twin
Project involvement
  • Smart buildings: participates in project discussions and exploration of intelligent applications.
  • Ship agents: contributes to agent projects for maritime applications.
  • Digital twins: contributes to innovation projects and project outputs, exploring applications of digital technologies.
Xiyuan Zhang

Xiyuan Zhang Xiyuan Zhang

Trainee Team Leader

2025 cohort · Software Engineering · Hubei University

Actively participates in PaddlePaddle training and technical exchange, developing algorithm and deployment skills through Cloud-Patrol and Smart Fish Tank. Shares innovation experience and explores AI applications through hands-on projects.

AI ApplicationsSoftware Engineering
Projects & awards
  • Cloud-Patrol: participated in the 2025 UP Camp project, receiving an Institute Third Prize.
  • Smart Fish Tank: participated in the 2026 summer camp project, receiving an Institute Grand Prize.
Jiayi Du

Jiayi Du Jiayi Du

People

2025 cohort · Software Engineering · Hubei University

Focuses on engineering AI algorithms and embedded systems through PaddlePaddle-based deep learning. Connects theory with hardware, refining architecture design, sensor control and software development to solve practical problems.

Deep LearningEmbedded SystemsSoftware Engineering
Projects & awards
  • Cloud-Patrol: designed the architecture for the 2025 UP Camp project, which received Third Prize.
  • Smart Fish Tank: developed sensor controls and software for the 2026 Institute summer camp project, which received Grand Prize.
Shenghan Xu

Shenghan Xu Shenghan Xu

People

2025 cohort · Software Engineering · Hubei University

Contributed to the “Greywater Reborn” project, which received the Grand Prize at the college summer camp, and is currently involved in preparing a project submission for the China International College Students’ Innovation Competition. Skilled in mechanical design using SOLIDWORKS, with hands-on experience in collaborative robotic arms and an interest in integrating mechanical structures with software control.

SOLIDWORKSMechanical DesignCollaborative Robotics

Team in action Life in the community

Hubei University students and teachers at the Baidu PaddlePaddle Wuhan AI Industry Empowerment Center in January 2025
UP Camp · Visiting the PaddlePaddle Wuhan AI Industry Empowerment CenterJanuary 2025 · Image source ↗

Interested in what we do?

Learn how to take part and start your first AI project.

How to join ↗